We have had the above issue and whilst I am not a Cisco SDWAN expert and not trying to resolve this at a vSmart / vManage / CEDGE level, I wanted to see if a potential workaround could be explored by exploring the options below... We have dual CEDGE boxes with MPLS & Biz Internet on each of them
a. Adjust the OAM DTLS keepalive and hold timers to reduce topology changes
b. Introduce interface dampening
c. Increase "Graceful Restart Timer" to the maximum
d. Route OAM tunnel via the MPLS as backup
Or all of the above or in any combinations as to upgrade the vManage & CEDGE in a fintech customer network is not simple as you can imagine.
